Random meta comment - but have you noticed that especially Codex generated websites tend to have these 01, 02, 03 numbers as part of card layouts? (see the section just below the header). It's almost the em-dash equivalent for AI built website detection.
One-shot websites from all the LLMs over the past 12 months (at the very least GPT, claude, glm, qwen based on my own experience) always add these stupid things. All you have to do is tell it to not include 2 digit section numbers in eyebrow text, and it'll stop instantly, but the people making these low effort sites just don't care.
I really despise these AI generated websites, they often look so samey. IDK about the numbering, but the overuse of cards everywhere, and abusing my favorite color schemes (tokyo-night, dracula style), really irritates me (esp, the color schemes, way to ruin even colors, guys).
It makes me dislike whatever they're trying to "sell" (literally or metaphorically). I wish people who wanted to be "creative" would you know... be creative, not just "push button, machine go brrr" which is what half this stuff spits out. Template, generic, no individuality. Hyper "professional" looking (but just "cool" enough to not be completely corporate).
On the other hand some of these look better and are more responsive than older sites that were definitely fuglier, klunky and less responsive. It feels like there's a middle ground, which I think is where good designer comes into play.
I don't think that's unique to AI --- many sites always had the same bland style, especially corporate ones, and I remember all the complaints about bootstrap.css when that was the trend many years ago.
Websites for what were ostensibly hobby projects didn't use to have a bland corporate look to them.
And it really gives me the ick when I see websites about things like game modding projects and they're full of AI linkedin-speak. I feel like if you were excited about building a mod for a game you wouldn't need AI to describe it for you, or write the announcements/reddit posts trying to reach out to the community.
I witness a large amount of AI generated pages at work and in the wild. Creative control belongs to the human because without it you get a scientifically accurate model without the stargazing factor.
I have built a few better sites after AI generates the shell. The colors and styling need about a dozen human interactions to get something passable as human created.
No, but... is the better answer in my mind. Learning to code is a considerable commitment. It takes years to get good enough to produce professional-grade software. If you extrapolate from the improvements we've seen in coding AI over just the last 12 months, this is just a bad allocation of your time.
Instead, as the article points out, learn to become a translator between the real world and AI code generation. Learn about industries that are relatively underserved by technology. Don't build tools for developers or engineers. Learn about construction, mining, waste management, oil and gas, manufacturing, logistics, government... then become the link between that industry and AI's ability to add value.
(Emphasis on relatively underserved — all of these have high-tech versions in some places, but the future isn't distributed evenly.)
> If you extrapolate from the improvements we've seen in coding AI over just the last 12 months
This is the achilles heel of your argument. I think people sometimes conflate realizing unmet potentials of LLMs with significant improvements, since there’s not been a fundamental change in how LLMs work as significant as the advancements we see in their application.
My thoughts too. Isn't much of the appeal of Apple to non-developers that it just sorta works without you having to think too much about it? That means that Apple necessarily just makes decisions on your behalf.
If they'd allow for more fine-grained control of low level aspects, a lot of their plug-and-play promise would no longer apply.
One room even features a secret entrance — hidden inside a rocket by the pool. — Ey — pshhh! :-)
I used to live in the inner city of Leipzig during my university days, and we had a door in the basement that led to wartime tunnels connecting it to other houses. It wasn't like a massive maze, just a hub with three other paths, each leading to another house nearby. From there, however, if they'd unlocked their doors, you could travel farther.
I was on a mission back then to befriend and convince random neighbours across the street to open their access doors and get some kind of network going, but I didn't meet with much enthusiasm.
I actually lived there for a few years, I guess I'd forgotten that the millenium bid fell through :)
Edit: By the by I do intend to visit again at some point if only to see the DEC exhibition at the town museum (if that's still on) and perhaps swing by Sweeney's pie shop (if that's still open).
Oh, and now I see from a sibling comment that there was another bid in 2022! It gives me a twinge of national pride that British rules on City-hood are so weird!
I just remember driving to the station once and just hating that whole area - loads of concrete. My experience may not apply to the rest of the place though, I grant you.
"In the end, the 2022 contest saw Reading beaten to the post by Dunfermline, Bangor, Douglas, Doncaster, Wrexham, Colchester, and nearby Milton Keynes. "
A lot of sources seem to cite Reading as 'England's largest town' but maybe that is by area, rather than population? Also depends on where you put the boundary.
That's where the problem lies. I spend too much of my life screaming into the void about UK addresses and what is and isn't a town :)
It's just about possible to create dataset of towns, cities, villages from mixing and matching and cutting up other geographies from the ONS (e.g. their Parished and Unparished data (though Unparished is its own nightmare), then the ONS Built-Up Areas dataset but... that one cuts out all the non-built parts so Milton Keynes is bunch of big blobs around the residential areas).
But if you frankenstein it, you lose all of the formal connections that define other links, and if you try to cheat via geospatial join, you will link police forces to areas they don't actually serve.
For Swindon, you're kinda stuck glueing together (going from memory here - please correct me if I'm off) the Parishes of Central Swindon, South Swindon and West Swindon. But these will overlap with some villages so it's not a true count.
And that's a natural moment to lead on to the next critical point: how're we measuring population? (I always get that Father Ted speech scene in my head thinking about this, start speaking - then twelve hours later "And now, we move on to liars")
Agreed, and it's a reasonable first reach for residential counts.
Problem is, it was last taken during 2021 - at the height of covid - and their estimates have been paying the price of it even since. Anything that touches on workday or employment they actively warn you away from, nowadays.
But then there's the elephant in the room: what if you're looking at somewhere like the City of London (the square mile one) where residential population is somewhere between 8,500 and 9,000 people?
EDIT: it slipped by me!
There is no single UK-wide census. NRS do Scotland's, last in 2022. NISRA handle Northern Ireland's census, last in 2021. And the ONS for England and Wales, last in 2021.
Well, "largest town" changes all the time anyway. Milton Keynes is "larger" than both by more than one metric, but it became a city in 2022.
The whole city status thing is a bit silly. It should definitely be based on more than population, but Milton Keynes getting it firmly established population is all it takes now.
The city is on a hill - or more precisely, on a ridge flanked by a river canyon on one side and a valley carved by a smaller stream on the other.
Due to that, the Znojmo catacombs are always above the water table & there was never really issues with drainage. There are a couple natural springs and overflowing wells in the catacombs & they are drained by the so-called Jesuit aqueduct, that collects all the water and drains it into the moat of the city fortification system. Nowadays that part is also underground & it just flows in the the Dyje river.
So during an emergency, the could get rid of the sewage with the aqueduct outflow & store solid was in some sections not in use I guess.
Leipzig is a bit more progressive/alternative, so I can see such a initiative to work out in the right place, but I doubt that today this is still a big thing? Lots of construction and modernisation going on in the city.
I love that you basically tried to start a neighborhood underground transit network and the only thing standing in your way was getting the neighbors to cooperate.
Quite easily, even retrospectively. It has no impact on his neighbours, doesn't create an eyesore, etc. He had the planning folks round, had them speak to his architect. It's all legit and done the correct way.
I appreciate that React & Co. have established it as a standard. I've used A-Frame lately, which lets you build WebVR components, and even there you declare your scene as
and I feel this is getting a bit much. Wouldn't it be better to have a JS API that lets you define objects and properties and compose your shader / scene etc. programmatically? (I've seen that Shaders offers that too — just asking in principle.)
Personally I don't mind it, but would say the main advantages offered by JSX for this kind of work (see: React Three Fiber) are more in its ease of composition and structuring larger projects.
Often in R3F projects you'll end up running per frame animations via the normal declarative property API in the useFrame hook anyway, as adjusting props in a loop is not great for performance. The JSX instead becomes a way to easily understand the hierarchy and nesting of objects, as well as quickly mount/unmount new elements.
It would be nice to (additionally) be able to load json in html without javascript (I am aware it is js object notation) for the same data. This way you could see it in a clean data structure and not serialized. Maybe using https://developer.mozilla.org/en-US/docs/Web/HTML/Reference/... or similar?
I do hobby work in Three.js and have always felt more comfortable using the declarative scene.add()-style API to manage my scene vs fiber / reactive stuff. But I imagine I'd feel differently if I were making 3d components tightly integrated with the state of the rest of a normal react app, rather than a fully immersive game scene.
I personally find it to be more readable, it gets rid of repeated syntax, adding children to parents, etc etc. But I think it's really a matter of personal preference.
